Published Paper Details:

Checking the quality of data in real time within event-driven processes guarantees that analytics and decisions based on AI are reliable. Standard batch methods for managing quality do not work well with huge and fast data streams. The review outlines a framework that depends on AI, includes rule-based validation, looks for anomalies using machine learning and can adapt based on changing needs to satisfy latency, scalability and robustness issues. Research studies have shown that using a hybrid approach with continuous drift detection gives high accuracy of 97% and few false positives. In a federated setting, data privacy is kept alongside effective threat detection. Among the key lessons is that model retraining is tough, available resources can be limited and privacy is essential, especially as it's important to balance accuracy and how quickly tasks can be completed. Because the framework is designed in modules, adding new features is simpler: simple checks are done by rules, but AI steps in when regular rules fail. If the drift-aware component recognizes changing data patterns, it leads to adaptive retraining or changes in the threshold, so the model does not lose performance. Further steps include deploying federated techniques on different kinds of edge devices, coming up with ways to describe data quality in understandable terms and adding governance policies to maintain compliance with future rules. Besides, having common benchmarking data and standards for performance is important for assessing real-time data quality solutions. It covers the various techniques in use, points out areas that still need work, like efficient resource use and integrating many types of data and details possible future projects for improving real-time data quality assurance. By gathering recent information, this article explains the present achievements and chances for new development to researchers and practitioners.